Import and process data

Learning

Model: Correct responses by age, trial, block number, and block condition

  correct_response_made
Predictors Odds Ratios SE
age group1 0.7626 0.0549
learning trial scaled 1.4490 0.0451
reward condition1 1.8834 0.1105
block number scaled 1.1543 0.0405
age group1 × learning
trial scaled
0.8999 0.0279
age group1 × reward
condition1
0.8880 0.0520
learning trial scaled ×
reward condition1
1.1825 0.0285
age group1 × block number
scaled
1.0190 0.0357
learning trial scaled ×
block number scaled
1.0280 0.0170
reward condition1 × block
number scaled
1.0688 0.0373
age group1 × learning
trial scaled × reward
condition1
0.9677 0.0232
(age group1 × learning
trial scaled) × block
number scaled
0.9944 0.0164
(age group1 × reward
condition1) × block
number scaled
1.0521 0.0367
(learning trial scaled ×
reward condition1) ×
block number scaled
0.9987 0.0176
(age group1 × learning
trial scaled × reward
condition1) × block
number scaled
1.0168 0.0179
Random Effects
σ2 3.29
τ00 subject_id 0.36
τ11 subject_id.re1.learning_trial_scaled 0.06
τ11 subject_id.re1.reward_condition1 0.23
τ11 subject_id.re1.block_number_scaled 0.08
τ11 subject_id.re1.learning_trial_scaled_by_reward_condition1 0.03
τ11 subject_id.re1.learning_trial_scaled_by_block_number_scaled 0.01
τ11 subject_id.re1.reward_condition1_by_block_number_scaled 0.07
τ11 subject_id.re1.learning_trial_scaled_by_reward_condition1_by_block_number_scaled 0.01
ρ01  
ρ01  
ICC 0.10
N subject_id 73
Observations 31536
Marginal R2 / Conditional R2 0.163 / 0.246

Figure 5A: Correct response by block condition, stimulus repetition, and age group

Supplementary Figure: Correct response by block condition and block number

Figure 5B: Generalization by block condition, category repetition, age group

Model: Correct response to first appearance of each stimulus

  correct_response_made
Predictors Odds Ratios SE
age group1 0.9285 0.0390
category rep scaled 1.1209 0.0403
reward condition1 1.4819 0.0673
block number scaled 1.0176 0.0368
age group1 × category rep
scaled
1.0128 0.0364
age group1 × reward
condition1
0.8858 0.0402
category rep scaled ×
reward condition1
1.2599 0.0454
age group1 × block number
scaled
0.9790 0.0354
category rep scaled ×
block number scaled
1.0071 0.0353
reward condition1 × block
number scaled
1.0774 0.0473
age group1 × category rep
scaled × reward
condition1
0.9448 0.0339
(age group1 × category
rep scaled) × block
number scaled
0.9514 0.0333
(age group1 × reward
condition1) × block
number scaled
1.0205 0.0448
(category rep scaled ×
reward condition1) ×
block number scaled
0.9952 0.0349
(age group1 × category
rep scaled × reward
condition1) × block
number scaled
1.0032 0.0352
Random Effects
σ2 3.29
τ00 subject_id 0.04
τ11 subject_id.re1.reward_condition1 0.06
τ11 subject_id.re1.block_number_scaled 0.01
τ11 subject_id.re1.reward_condition1_by_block_number_scaled 0.05
ρ01  
ρ01  
ICC 0.01
N subject_id 73
Observations 3942
Marginal R2 / Conditional R2 0.073 / 0.085

Model: Category win-stay lose-shift

  WSLS
Predictors Odds Ratios SE
age group1 0.9060 0.0294
learning trial scaled 1.0102 0.0146
reward condition1 1.5157 0.0507
block number scaled 1.0603 0.0218
age group1 × learning
trial scaled
0.9983 0.0144
age group1 × reward
condition1
0.8968 0.0300
learning trial scaled ×
reward condition1
1.1446 0.0188
age group1 × block number
scaled
1.0197 0.0210
learning trial scaled ×
block number scaled
0.9811 0.0132
reward condition1 × block
number scaled
1.0701 0.0193
age group1 × learning
trial scaled × reward
condition1
0.9802 0.0161
(age group1 × learning
trial scaled) × block
number scaled
1.0132 0.0136
(age group1 × reward
condition1) × block
number scaled
1.0053 0.0181
(learning trial scaled ×
reward condition1) ×
block number scaled
1.0040 0.0137
(age group1 × learning
trial scaled × reward
condition1) × block
number scaled
1.0141 0.0138
Random Effects
σ2 3.29
τ00 subject_id 0.06
τ11 subject_id.re1.learning_trial_scaled 0.00
τ11 subject_id.re1.reward_condition1 0.07
τ11 subject_id.re1.block_number_scaled 0.02
τ11 subject_id.re1.learning_trial_scaled_by_reward_condition1 0.01
τ11 subject_id.re1.learning_trial_scaled_by_block_number_scaled 0.00
τ11 subject_id.re1.reward_condition1_by_block_number_scaled 0.01
τ11 subject_id.re1.learning_trial_scaled_by_reward_condition1_by_block_number_scaled 0.00
ρ01  
ρ01  
ICC 0.02
N subject_id 73
Observations 29799
Marginal R2 / Conditional R2 0.064 / 0.082

Supplementary Figure: WSLS by age group

Supplementary Figure: WSLS by block number

Memory

Overall descriptive memory stats

mean_mem se_mem
0.7579 0.01331

Memory delay stats

mean_delay sd_delay min_delay max_delay
7.123 1.092 5 10
age_group mean_delay sd_delay min_delay max_delay
Children 7.118 1.122 6 10
Adults 7.128 1.08 5 9
memory_delay age_group N
5 Adults 1
6 Children 13
6 Adults 13
7 Children 9
7 Adults 9
8 Children 8
8 Adults 12
9 Children 3
9 Adults 4
10 Children 1

Figure 5D: AUC values by age group, memory specificity, block condition

Model: AUCs by age group, reward condition, memory specificity

  AUC
Predictors Estimates SE
age group1 -0.0029 0.0143
reward condition1 -0.0179 0.0040
foil type1 0.0428 0.0040
age group1 × reward
condition1
-0.0006 0.0040
age group1 × foil type1 0.0036 0.0040
reward condition1 × foil
type1
0.0078 0.0040
age group1 × reward
condition1 × foil type1
0.0031 0.0040
Random Effects
σ2 0.00
τ00 subject_id 0.01
ICC 0.74
N subject_id 73
Observations 292
Marginal R2 / Conditional R2 0.108 / 0.772

Model: AUCs by age, reward condition, foil type, delay

  AUC
Predictors Estimates SE
age group1 -0.0029 0.0145
reward condition1 -0.0180 0.0040
foil type1 0.0428 0.0040
mem delay scaled 0.0035 0.0145
age group1 × reward
condition1
-0.0006 0.0040
age group1 × foil type1 0.0037 0.0040
reward condition1 × foil
type1
0.0078 0.0040
age group1 × mem delay
scaled
0.0037 0.0145
reward condition1 × mem
delay scaled
-0.0001 0.0040
foil type1 × mem delay
scaled
0.0097 0.0040
age group1 × reward
condition1 × foil type1
0.0031 0.0040
(age group1 × reward
condition1) × mem delay
scaled
-0.0065 0.0040
(age group1 × foil type1)
× mem delay scaled
0.0032 0.0040
(reward condition1 × foil
type1) × mem delay scaled
-0.0001 0.0040
(age group1 × reward
condition1 × foil type1)
× mem delay scaled
0.0021 0.0040
Random Effects
σ2 0.00
τ00 subject_id 0.01
ICC 0.75
N subject_id 73
Observations 292
Marginal R2 / Conditional R2 0.114 / 0.782

Supplementary Figure: AUC values with delay

RL modeling

Choice weights

Figure: Distribution of choice weights

Figure 5C: Choice weights box plot

Model: Choice weights by block condition and age group

  est
Predictors Estimates SE
abstraction1 -0.0404 0.0547
reward condition1 0.0417 0.0547
age group1 -0.0842 0.0980
abstraction1 × reward
condition1
0.1356 0.0547
abstraction1 × age group1 0.0967 0.0547
reward condition1 × age
group1
0.0157 0.0547
abstraction1 × reward
condition1 × age group1
-0.0347 0.0547
Random Effects
σ2 0.87
τ00 subject_id 0.48
ICC 0.36
N subject_id 73
Observations 292
Marginal R2 / Conditional R2 0.029 / 0.375

Model: Exemplar choice weights by condition

  est
Predictors Estimates SE
reward condition1 -0.0974 0.0606
Random Effects
σ2 0.54
τ00 subject_id 0.87
ICC 0.62
N subject_id 73
Observations 146
Marginal R2 / Conditional R2 0.007 / 0.622

Model: Category choice weights by condition

  est
Predictors Estimates SE
reward condition1 0.1785 0.0642
Random Effects
σ2 0.60
τ00 subject_id 0.69
ICC 0.54
N subject_id 73
Observations 146
Marginal R2 / Conditional R2 0.024 / 0.546

Model: Relation between exemplar and category choice weights

  beta_e_scaled
Predictors Estimates SE
beta c scaled -0.0181 0.0798
age scaled 0.1214 0.1099
block condition1 -0.0752 0.0496
beta c scaled × age
scaled
-0.1044 0.0824
beta c scaled × block
condition1
-0.1448 0.0545
age scaled × block
condition1
-0.0146 0.0512
beta c scaled × age
scaled × block condition1
0.0345 0.0547
Random Effects
σ2 0.33
τ00 subject_id 0.69
ICC 0.67
N subject_id 73
Marginal R2 / Conditional R2 0.057 / 0.692
## 
## Call:
## lm(formula = beta_e_scaled ~ beta_c_scaled * age_scaled, data = beta_ests_wide_exemp_pred)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -2.4370 -0.4634  0.1487  0.6339  1.7488 
## 
## Coefficients:
##                            Estimate Std. Error t value Pr(>|t|)  
## (Intercept)              -0.0002356  0.1131188  -0.002    0.998  
## beta_c_scaled             0.2815002  0.1150579   2.447    0.017 *
## age_scaled                0.1871373  0.1156527   1.618    0.110  
## beta_c_scaled:age_scaled -0.0038982  0.1173137  -0.033    0.974  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.9646 on 69 degrees of freedom
## Multiple R-squared:  0.1083, Adjusted R-squared:  0.06957 
## F-statistic: 2.795 on 3 and 69 DF,  p-value: 0.04666
## 
## Call:
## lm(formula = beta_e_scaled ~ beta_c_scaled * age_scaled, data = beta_ests_wide_cat_pred)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -2.19715 -0.79074  0.04502  0.83269  1.78719 
## 
## Coefficients:
##                            Estimate Std. Error t value Pr(>|t|)
## (Intercept)               3.404e-05  1.182e-01   0.000    1.000
## beta_c_scaled             1.039e-01  1.196e-01   0.869    0.388
## age_scaled                1.021e-01  1.197e-01   0.853    0.397
## beta_c_scaled:age_scaled -2.219e-02  1.210e-01  -0.183    0.855
## 
## Residual standard error: 1.01 on 69 degrees of freedom
## Multiple R-squared:  0.02178,    Adjusted R-squared:  -0.02075 
## F-statistic: 0.5121 on 3 and 69 DF,  p-value: 0.6753

Relations between choice weights and points earned

## Mixed Model Anova Table (Type 3 tests, S-method)
## 
## Model: total_points ~ age_group * beta_scaled * abstraction * reward_condition + 
## Model:     (1 | subject_id)
## Data: beta_ests_points
##                                                Effect        df          F
## 1                                           age_group  1, 60.48  15.36 ***
## 2                                         beta_scaled 1, 267.98   10.74 **
## 3                                         abstraction 1, 195.18       0.69
## 4                                    reward_condition 1, 194.54 451.24 ***
## 5                               age_group:beta_scaled 1, 267.98       0.27
## 6                               age_group:abstraction 1, 195.18       0.00
## 7                             beta_scaled:abstraction 1, 227.31       0.38
## 8                          age_group:reward_condition 1, 194.54     4.69 *
## 9                        beta_scaled:reward_condition 1, 203.34  31.13 ***
## 10                       abstraction:reward_condition 1, 195.93       0.09
## 11                  age_group:beta_scaled:abstraction 1, 227.31       0.23
## 12             age_group:beta_scaled:reward_condition 1, 203.34       0.39
## 13             age_group:abstraction:reward_condition 1, 195.93       0.00
## 14           beta_scaled:abstraction:reward_condition 1, 208.19  30.89 ***
## 15 age_group:beta_scaled:abstraction:reward_condition 1, 208.19       1.49
##    p.value
## 1    <.001
## 2     .001
## 3     .407
## 4    <.001
## 5     .604
## 6     .946
## 7     .538
## 8     .032
## 9    <.001
## 10    .762
## 11    .631
## 12    .532
## 13    .979
## 14   <.001
## 15    .224
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '+' 0.1 ' ' 1
## 
## Call:
## lm(formula = total_points ~ est, data = beta_ests_points_c_c)
## 
## Residuals:
##    Min     1Q Median     3Q    Max 
## -93.64 -36.87   3.86  39.43  83.93 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)   36.795      5.880   6.257 2.62e-08 ***
## est           39.829      4.692   8.489 2.07e-12 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 46.65 on 71 degrees of freedom
## Multiple R-squared:  0.5037, Adjusted R-squared:  0.4968 
## F-statistic: 72.07 on 1 and 71 DF,  p-value: 2.069e-12
## 
## Call:
## lm(formula = total_points ~ est, data = beta_ests_points_e_c)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -131.981  -52.267   -2.401   41.211  143.996 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)   50.653      7.626   6.642 5.28e-09 ***
## est           16.470      6.364   2.588   0.0117 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 63.31 on 71 degrees of freedom
## Multiple R-squared:  0.0862, Adjusted R-squared:  0.07333 
## F-statistic: 6.698 on 1 and 71 DF,  p-value: 0.0117
## 
## Call:
## lm(formula = total_points ~ est, data = beta_ests_points_c_e)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -64.897 -23.138  -3.735  21.206  87.594 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  -24.448      4.025  -6.074 5.56e-08 ***
## est           -6.766      3.665  -1.846    0.069 .  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 34.22 on 71 degrees of freedom
## Multiple R-squared:  0.04581,    Adjusted R-squared:  0.03237 
## F-statistic: 3.408 on 1 and 71 DF,  p-value: 0.06904
## 
## Call:
## lm(formula = total_points ~ est, data = beta_ests_points_e_e)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -63.454 -13.244  -2.002  14.115  72.974 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  -34.370      3.306 -10.395 6.60e-16 ***
## est           19.236      2.576   7.468 1.62e-10 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 26.22 on 71 degrees of freedom
## Multiple R-squared:  0.4399, Adjusted R-squared:  0.432 
## F-statistic: 55.77 on 1 and 71 DF,  p-value: 1.616e-10

Supplementary Figure: Points earned vs. beta values

Relations between learning and memory

Do points earned during learning relate to memory?

Model: AUC by points earned, reward condition, age group

  AUC
Predictors Estimates SE
age group1 0.0000 0.0143
points scaled 0.0403 0.0104
abstraction1 0.0316 0.0061
reward condition1 -0.0397 0.0080
age group1 × points
scaled
0.0011 0.0104
age group1 × abstraction1 0.0020 0.0061
points scaled ×
abstraction1
-0.0030 0.0069
age group1 × reward
condition1
-0.0021 0.0080
points scaled × reward
condition1
-0.0135 0.0086
abstraction1 × reward
condition1
0.0098 0.0061
age group1 × points
scaled × abstraction1
-0.0112 0.0069
age group1 × points
scaled × reward
condition1
0.0108 0.0086
age group1 × abstraction1
× reward condition1
0.0133 0.0061
points scaled ×
abstraction1 × reward
condition1
0.0139 0.0069
age group1 × points
scaled × abstraction1 ×
reward condition1
0.0024 0.0069
Random Effects
σ2 0.00
τ00 subject_id 0.01
ICC 0.71
N subject_id 73
Observations 292
Marginal R2 / Conditional R2 0.164 / 0.754

Figure 5E: AUC by performance group and reward condition

Do choice weights relate to memory?

Model: AUC by age group, exemplar choice weights, specificity, block condition

  AUC
Predictors Estimates SE
age group1 0.0027 0.0123
beta scaled 0.0345 0.0079
abstraction1 0.0431 0.0041
reward condition1 -0.0136 0.0042
age group1 × beta scaled -0.0159 0.0079
age group1 × abstraction1 0.0039 0.0041
beta scaled ×
abstraction1
0.0015 0.0042
age group1 × reward
condition1
-0.0035 0.0042
beta scaled × reward
condition1
-0.0018 0.0046
abstraction1 × reward
condition1
0.0079 0.0041
age group1 × beta scaled
× abstraction1
0.0020 0.0042
age group1 × beta scaled
× reward condition1
0.0061 0.0046
age group1 × abstraction1
× reward condition1
0.0034 0.0041
beta scaled ×
abstraction1 × reward
condition1
0.0010 0.0042
age group1 × beta scaled
× abstraction1 × reward
condition1
0.0008 0.0042
Random Effects
σ2 0.00
τ00 subject_id 0.01
ICC 0.66
N subject_id 73
Observations 292
Marginal R2 / Conditional R2 0.203 / 0.733

Figure 5F: AUC by exemplar choice weights: model effects

Model: AUC by age group, category choice weights, specificity, block condition

  AUC
Predictors Estimates SE
age group1 -0.0048 0.0144
beta scaled 0.0083 0.0075
abstraction1 0.0423 0.0040
reward condition1 -0.0192 0.0041
age group1 × beta scaled 0.0094 0.0075
age group1 × abstraction1 0.0040 0.0040
beta scaled ×
abstraction1
0.0111 0.0040
age group1 × reward
condition1
-0.0019 0.0041
beta scaled × reward
condition1
0.0011 0.0045
abstraction1 × reward
condition1
0.0061 0.0040
age group1 × beta scaled
× abstraction1
-0.0013 0.0040
age group1 × beta scaled
× reward condition1
0.0115 0.0045
age group1 × abstraction1
× reward condition1
0.0035 0.0040
beta scaled ×
abstraction1 × reward
condition1
0.0025 0.0040
age group1 × beta scaled
× abstraction1 × reward
condition1
-0.0031 0.0040
Random Effects
σ2 0.00
τ00 subject_id 0.01
ICC 0.76
N subject_id 73
Observations 292
Marginal R2 / Conditional R2 0.126 / 0.789

Figure 5F: AUC by category choice weights: model effects

Relations between age group and other model parameters

Model: Alpha choice values by age group

## 
## Call:
## lm(formula = alpha ~ age_group, data = param_ests)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -2.9662 -1.6183 -0.4774  1.8037  5.2505 
## 
## Coefficients:
##                 Estimate Std. Error t value Pr(>|t|)    
## (Intercept)      -3.6116     0.3407 -10.600 2.82e-16 ***
## age_groupAdults   0.3878     0.4661   0.832    0.408    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 1.987 on 71 degrees of freedom
## Multiple R-squared:  0.009655,   Adjusted R-squared:  -0.004293 
## F-statistic: 0.6922 on 1 and 71 DF,  p-value: 0.4082

Model: Alpha cf values by age group

## 
## Call:
## lm(formula = alpha_cf ~ age_group, data = param_ests)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -6.2676 -0.5954  0.4327  1.1856  4.0781 
## 
## Coefficients:
##                 Estimate Std. Error t value Pr(>|t|)    
## (Intercept)      -1.9938     0.3326  -5.995 7.69e-08 ***
## age_groupAdults   0.1240     0.4550   0.273    0.786    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 1.939 on 71 degrees of freedom
## Multiple R-squared:  0.001045,   Adjusted R-squared:  -0.01302 
## F-statistic: 0.07431 on 1 and 71 DF,  p-value: 0.786

Model: Initial Q by age

## 
## Call:
## lm(formula = q_init ~ age_group, data = param_ests)
## 
## Residuals:
##    Min     1Q Median     3Q    Max 
## -6.895 -2.615  1.736  2.309  2.962 
## 
## Coefficients:
##                 Estimate Std. Error t value Pr(>|t|)   
## (Intercept)       1.4896     0.5168   2.882  0.00522 **
## age_groupAdults  -0.2880     0.7071  -0.407  0.68497   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 3.014 on 71 degrees of freedom
## Multiple R-squared:  0.002332,   Adjusted R-squared:  -0.01172 
## F-statistic: 0.1659 on 1 and 71 DF,  p-value: 0.685